11488/11688/11788 — Spring 2021
Grading will be based on 2 major components: Weekly quizzes and Bi-weekly assignments.
The course grading will be based on a total of 100 points:
| 💻 | Assignments 80 points |
These assignments will be released tri-weekly, and will be 20 points each. |
| 📑 | Quizzes 20 points |
There will be about 8-12 quizzes in all. We will drop your two lowest quizzes. Each quiz will be worth equal points, and based on the lectures of the past week. |
The four assignments for this course will be based on state-of-the-art techniques to track cyber-crimes. A brief description of them is as follows:
The first part of this assignment will be based on getting familiar with “packet sniffing” and analyzing data network characteristics using the network packet trace This assignment will be based on getting familiar with network and deep web forensics. There are six parts to this homework. The first three parts will be based on “packet sniffing” and analyzing data network characteristics using the network packet trace (or pcap), and understanding to identify different network threats and vulnerabilities. The last three parts with be based on getting familiar with the dark web by analyzing the characteristics of the Tor circuit, crawling and modeling topics around the dark web, and finally getting familiar with the Tor hidden services. The six parts are briefly described below:
| Part 1: Wireshark Warm-up | In this part, you will manually analyze the network packet trace for BitTorrent traffic using Wireshark. |
| Part 2: Exploring Network Threats & Vulnerabilities | In this part you will examine the network packet trace for specific vulnerabilities using Wireshark. |
| Part 3: Identifying SYN-ners | In this part, you will automate the process of packet sniffing by writing a program to analyze a pcap file to detect “port-scanning”. |
| Part 4: Tor Circuit Analysis | In this part, you will use your knowledge of the Tor circuits to formulate a problem, plan a solution/analysis, collect data, and analyze the data to characterize the Tor circuit. |
| Part 5: Crawling the Dark Web | In this part, you will crawl the dark web to identify suspicious activity. |
| Part 6: Using Tor Hidden Services | Finally, you will create a site on the dark web using Tor Hidden services, and populate it with your solution from previous parts for us to grade. |
In this assignment, you will apply your understanding of artificial intelligence to build knowledge and rules for forensic applications, and then use your knowledge of machine learning to engineer features, and build models for social media forensics. This assignment has 4 parts which are as follows:
| Part 1: Building a COVID-net | In this part, you will create a semantic network to create knowledge representation for a problem related to COVID-19. |
| Part 2: Intrusion Detection Using Rules and Exceptions | In this part, you will extract decision rules to classify types of intrusion from network data using the algorithms taught in the class. |
| Part 3: Conducting A Forensic Background Check | In this part, you will conduct a forensic background check of a person using social media and text forensic tools, and create a detailed report. |
| Part 4: COVID-19 Twitter Misinformation Analysis | In this problem you will engineer features and create models that distinguish real information from misinformation about COVID-19 on Twitter. |
This assignment will be based on machine learning, audio/voice, and image forensics. There are three parts to this homework
| Part 1: Analyzing Audios for COVID-19 signatures | In this part, you will collect and analyze audios (using PRAAT) to identify features to detect COVID-19 signatures in voice. |
| Part 2: Automatic Speaker Age Estimation | In this part, you will use machine learning algorithms to build and optimize models for estimating age of people from their voice recordings. |
| Part 3: Identifying Image Rescaling | In this part, you will explore how to detect if an image has been tampered computationally using correlation analysis. |
DeepFake is the process of swapping the face of a source identity with a target identity in a given video/image using deep learning techniques. The name originates from “deep learning” and “fake”. There are many technologies that can used to generate fakes but ones that use deep learning are called DeepFakes and tend to be the most photo-realistic. This assignment is to familiarize you with “DeepFakes”. There are two parts to this assignment which are as follows:
| Part 1: Generating DeepFakes | In this part, you will generate your own DeepFake using DeepFaceLab. |
| Part 2: Detecting DeepFakes | In this part, you will create a neural network based model to detect DeepFakes. |
There’s a maximum score: 100 points.